How to use the Azure Resource Visualizer agent skill

What is an agent skill and what does this one do?

An agent skill is a folder of instructions and supporting files that an AI agent loads when a task matches the skill's description. This one lives in the SKILL.md file and is triggered when you ask for a diagram of your Azure resources or for help understanding how those resources relate to each other.

Once active, the agent lists the resources in a resource group and captures each one's name, type, SKU or tier, location, key settings, network details, identities, and connections. It maps relationships such as network connections, data flows, managed identity use, configuration links like app settings pointing to Key Vaults, and parent-child dependencies.

• Network connections: VNet peering, subnet assignments, NSG rules, private endpoints
• Data flow: for example apps to databases, or functions to storage
• Identity: managed identities connecting to resources
• Configuration: app settings pointing to Key Vaults, connection strings
• Dependencies: parent-child relationships and required resources

How do you install the Azure Resource Visualizer skill?

Download the ZIP of the skill from this page and unzip it. You can install it in one of two ways. For a personal skill available across your projects, place the unzipped folder in ~/.claude/skills/. For a skill that applies to one project only, place it in that project's .claude/skills/ folder.

If you use the Claude apps instead of files on disk, upload the skill under Settings → Capabilities → Skills.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into the project's .claude/skills/ folder
• Claude apps: upload under Settings → Capabilities → Skills

What do you need before you start?

The skill reads your Azure environment, so it needs a way to query Azure. The source names two options: the Azure CLI (az) or Azure MCP tools. You also need an authenticated Azure account with read access to the resource groups you want to diagram. Without credentials the skill cannot read anything.

A Mermaid-capable viewer is optional. The diagram is written as Mermaid code inside a markdown file, so a viewer that renders Mermaid will show it as a picture.

• Azure CLI (az) or Azure MCP tools
• An authenticated Azure account with read access to the target resource groups
• A Mermaid-capable viewer to render the diagram (optional)

How do you trigger the skill with a natural request?

You do not need special syntax. Describe what you want in plain language and mention your resource group if you know its name. If you do not name one, the agent lists your resource groups with their locations and waits for you to pick one by number or name before it continues.

Good requests say what you want to see and why, for example a diagram of your resources or help understanding how they fit together.

• Draw a diagram of the resources in my resource group rg-web-prod.
• Help me understand how the resources in my Azure account relate to each other.
• Show me the architecture of my Azure resource group and what connects to the database.
• Map the network and identity links between my resources.

What output should you expect?

The agent writes a markdown file named after the resource group, in the form [resource-group-name]-architecture.md. It places the file in the workspace root or in a docs/ folder if one exists. The file follows the layout of the included template: a header with the resource group, subscription and region, a summary, a resource inventory table, the Mermaid architecture diagram, relationship details for network, data flow, identity and dependencies, and a notes section with observations and possible issues.

The diagram groups resources into subgraphs by layer, such as network, compute, data and security, and labels connections with what flows across them. The source notes that very large groups of 50 or more resources may be split into several diagrams by layer.

What are the safety notes?

The skill runs scripts and uses the network, because it queries Azure to read your resources. The source describes the work as read-only analysis: the agent is told not to modify or delete Azure resources.

The bigger risk is the output. The generated markdown file may include resource names, IP ranges and settings. If you share it or commit it to a repository, review it first and remove anything sensitive. Do not use the skill on a resource group that holds configuration you should not expose.

• The skill is designed for read-only analysis and should not change Azure resources.
• Review the generated file for names, IP address ranges and settings before sharing or committing it.
• Check that your Azure account only has the read access you intend the agent to use.

What tips make the results better?

Name the resource group in your request to skip the selection step. Ask for a layer-by-layer view if the group is large. Check the notes section, since it lists potential issues the agent noticed, but verify them against your own knowledge before acting on them.

If the agent reports no resources or a permission problem, the source says to verify the resource group name and check RBAC for the missing access.

• Name the resource group up front to avoid the selection prompt.
• For large groups, ask for separate diagrams per layer.
• If something is missing from the diagram, ask the agent to recheck the resources and relationships it could not confirm.
• Use a Mermaid viewer to render the diagram and preview it before sharing.